Hugging Face's logo

ayushexel
/
colbert-ModernBERT-base-2-neg-4-epoch-gooaq-1995000

Sentence Similarity
Safetensors
sentence-transformers
PyLate
modernbert
ColBERT
feature-extraction
Generated from Trainer
dataset_size:3787118
loss:Contrastive
Eval Results
Model card Files Community
colbert-ModernBERT-base-2-neg-4-epoch-gooaq-1995000 / README.md
ayushexel's picture
ayushexel
Add new SentenceTransformer model
900ad26 verified
preview code
|
raw
history blame contribute delete
55.1 kB
metadata
tags:
  - ColBERT
  - PyLate
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:3787118
  - loss:Contrastive
base_model: answerdotai/ModernBERT-base
pipeline_tag: sentence-similarity
library_name: PyLate
metrics:
  - accuracy
model-index:
  - name: PyLate model based on answerdotai/ModernBERT-base
    results:
      - task:
          type: col-berttriplet
          name: Col BERTTriplet
        dataset:
          name: Unknown
          type: unknown
        metrics:
          - type: accuracy
            value: 0.5331999659538269
            name: Accuracy

PyLate model based on answerdotai/ModernBERT-base

This is a PyLate model finetuned from answerdotai/ModernBERT-base. It maps sentences & paragraphs to sequences of 128-dimensional dense vectors and can be used for semantic textual similarity using the MaxSim operator.

Model Details

Model Description

  • Model Type: PyLate model
  • Base model: answerdotai/ModernBERT-base
  • Document Length: 180 tokens
  • Query Length: 32 tokens
  • Output Dimensionality: 128 tokens
  • Similarity Function: MaxSim

Model Sources

Full Model Architecture 

ColBERT(
  (0): Transformer({'max_seq_length': 179, 'do_lower_case': False}) with Transformer model: ModernBertModel 
  (1): Dense({'in_features': 768, 'out_features': 128, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity'})
)

Usage

First install the PyLate library:

pip install -U pylate

Retrieval

PyLate provides a streamlined interface to index and retrieve documents using ColBERT models. The index leverages the Voyager HNSW index to efficiently handle document embeddings and enable fast retrieval.

Indexing documents

First, load the ColBERT model and initialize the Voyager index, then encode and index your documents:

from pylate import indexes, models, retrieve

# Step 1: Load the ColBERT model
model = models.ColBERT(
    model_name_or_path=pylate_model_id,
)

# Step 2: Initialize the Voyager index
index = indexes.Voyager(
    index_folder="pylate-index",
    index_name="index",
    override=True,  # This overwrites the existing index if any
)

# Step 3: Encode the documents
documents_ids = ["1", "2", "3"]
documents = ["document 1 text", "document 2 text", "document 3 text"]

documents_embeddings = model.encode(
    documents,
    batch_size=32,
    is_query=False,  # Ensure that it is set to False to indicate that these are documents, not queries
    show_progress_bar=True,
)

# Step 4: Add document embeddings to the index by providing embeddings and corresponding ids
index.add_documents(
    documents_ids=documents_ids,
    documents_embeddings=documents_embeddings,
)

Note that you do not have to recreate the index and encode the documents every time. Once you have created an index and added the documents, you can re-use the index later by loading it:

# To load an index, simply instantiate it with the correct folder/name and without overriding it
index = indexes.Voyager(
    index_folder="pylate-index",
    index_name="index",
)

Retrieving top-k documents for queries

Once the documents are indexed, you can retrieve the top-k most relevant documents for a given set of queries. To do so, initialize the ColBERT retriever with the index you want to search in, encode the queries and then retrieve the top-k documents to get the top matches ids and relevance scores:

# Step 1: Initialize the ColBERT retriever
retriever = retrieve.ColBERT(index=index)

# Step 2: Encode the queries
queries_embeddings = model.encode(
    ["query for document 3", "query for document 1"],
    batch_size=32,
    is_query=True,  #  # Ensure that it is set to False to indicate that these are queries
    show_progress_bar=True,
)

# Step 3: Retrieve top-k documents
scores = retriever.retrieve(
    queries_embeddings=queries_embeddings,
    k=10,  # Retrieve the top 10 matches for each query
)

Reranking

If you only want to use the ColBERT model to perform reranking on top of your first-stage retrieval pipeline without building an index, you can simply use rank function and pass the queries and documents to rerank:

from pylate import rank, models

queries = [
    "query A",
    "query B",
]

documents = [
    ["document A", "document B"],
    ["document 1", "document C", "document B"],
]

documents_ids = [
    [1, 2],
    [1, 3, 2],
]

model = models.ColBERT(
    model_name_or_path=pylate_model_id,
)

queries_embeddings = model.encode(
    queries,
    is_query=True,
)

documents_embeddings = model.encode(
    documents,
    is_query=False,
)

reranked_documents = rank.rerank(
    documents_ids=documents_ids,
    queries_embeddings=queries_embeddings,
    documents_embeddings=documents_embeddings,
)

Evaluation

Metrics

Col BERTTriplet

  • Evaluated with pylate.evaluation.colbert_triplet.ColBERTTripletEvaluator
Metric Value
accuracy 0.5332

Training Details

Training Dataset

Unnamed Dataset

  • Size: 3,787,118 training samples
  • Columns: question, answer, and negative
  • Approximate statistics based on the first 1000 samples:
    question answer negative
    type string string string
    details
    • min: 9 tokens
    • mean: 13.17 tokens
    • max: 21 tokens
    • min: 17 tokens
    • mean: 31.79 tokens
    • max: 32 tokens
    • min: 15 tokens
    • mean: 31.69 tokens
    • max: 32 tokens
  • Samples:
    question answer negative
    what does it mean when you dream you are a vampire? Dream Bible - Dream Interpretation of Vampires. To see a vampire in your dream symbolizes an aspect of your personality that is parasitic or selfishly feeds off others. A person or situation that drains you of time, energy, or resources. ... To dream of being a vampire represents a selfish need to use or feed off others. Many people believe that witches, warlocks, vampires, werewolves are necessarily seen as a sign of something bad. ... If you dream that a vampire is chasing you, but it is not able to catch you, it means that you are afraid of someone, but your fear is groundless, this person will not hurt you.
    what does it mean when you dream you are a vampire? Dream Bible - Dream Interpretation of Vampires. To see a vampire in your dream symbolizes an aspect of your personality that is parasitic or selfishly feeds off others. A person or situation that drains you of time, energy, or resources. ... To dream of being a vampire represents a selfish need to use or feed off others. Blood on the ground in a dream indicates that the dreamer should be careful of unusual or new friendships. ... To lose blood in your dream represents that you may be tired in your waking life and that you feel emotional. If you dream you are in the hospital and you see blood, it means that past actions may haunt you.
    what is the difference between thrombosis and atherosclerosis? Arterial thrombosis usually affects people whose arteries are clogged with fatty deposits. This is known as atherosclerosis. These deposits cause the arteries to harden and narrow over time and increase the risk of blood clots. Advertisement. Atherosclerosis is a specific type of arteriosclerosis, but the terms are sometimes used interchangeably. Atherosclerosis refers to the buildup of fats, cholesterol and other substances in and on your artery walls (plaque), which can restrict blood flow.
  • Loss: pylate.losses.contrastive.Contrastive

Evaluation Dataset

Unnamed Dataset

  • Size: 5,000 evaluation samples
  • Columns: question, answer, and negative_1
  • Approximate statistics based on the first 1000 samples:
    question answer negative_1
    type string string string
    details
    • min: 9 tokens
    • mean: 12.96 tokens
    • max: 24 tokens
    • min: 16 tokens
    • mean: 31.73 tokens
    • max: 32 tokens
    • min: 14 tokens
    • mean: 31.43 tokens
    • max: 32 tokens
  • Samples:
    question answer negative_1
    are upc codes and barcodes the same? The UPC-A barcode was the original format for product barcodes. ... The only major difference is the placement of the numbers below (human readable numbers) which are there only as a back-up in case the barcode doesn't scan properly and the information has to be manually entered into the point of sale system. Barcodes do not support any of the mentioned attributes. Thus, QR Codes will not be replaced by barcodes.
    what does it mean when it says your application is under review? "Under review" is a phrase that typically means your application is being screened by human resources or the hiring manager. "Applicants being selected" indicates that hiring managers are selecting candidates for interviews. "Referred to hiring manager" means your application has passed initial HR screening. There are two different types of review: In Review and Under Review. If your verification status is In Review, this means that your ID uploaded successfully and your submission is being reviewed automatically. Automated reviews typically take anywhere from 5 minutes to 2 hours.
    when was the last time the kansas city chiefs won a playoff game? The Chiefs were overwhelmed by the Bills and lost the game by a score of 30–13. The Chiefs' victory on January 16, 1994, against the Oilers remained the franchise's last post-season victory for 21 years until their 30–0 victory over the Houston Texans on January 9, 2016. The Broncos have not beaten the Chiefs since a Week 2 game in Kansas City in 2015. Kansas City's streak is the second longest in the series' history. Only the Chiefs' 11-game winning streak from 1964 to 1969 lasted longer.
  • Loss: pylate.losses.contrastive.Contrastive

Training Hyperparameters

Non-Default Hyperparameters

  • eval_strategy: steps
  • per_device_train_batch_size: 128
  • per_device_eval_batch_size: 128
  • learning_rate: 3e-06
  • num_train_epochs: 5
  • warmup_ratio: 0.1
  • seed: 12
  • bf16: True
  • dataloader_num_workers: 12
  • load_best_model_at_end: True

All Hyperparameters

Click to expand
  • overwrite_output_dir: False
  • do_predict: False
  • eval_strategy: steps
  • prediction_loss_only: True
  • per_device_train_batch_size: 128
  • per_device_eval_batch_size: 128
  • per_gpu_train_batch_size: None
  • per_gpu_eval_batch_size: None
  • gradient_accumulation_steps: 1
  • eval_accumulation_steps: None
  • torch_empty_cache_steps: None
  • learning_rate: 3e-06
  • weight_decay: 0.0
  • adam_beta1: 0.9
  • adam_beta2: 0.999
  • adam_epsilon: 1e-08
  • max_grad_norm: 1.0
  • num_train_epochs: 5
  • max_steps: -1
  • lr_scheduler_type: linear
  • lr_scheduler_kwargs: {}
  • warmup_ratio: 0.1
  • warmup_steps: 0
  • log_level: passive
  • log_level_replica: warning
  • log_on_each_node: True
  • logging_nan_inf_filter: True
  • save_safetensors: True
  • save_on_each_node: False
  • save_only_model: False
  • restore_callback_states_from_checkpoint: False
  • no_cuda: False
  • use_cpu: False
  • use_mps_device: False
  • seed: 12
  • data_seed: None
  • jit_mode_eval: False
  • use_ipex: False
  • bf16: True
  • fp16: False
  • fp16_opt_level: O1
  • half_precision_backend: auto
  • bf16_full_eval: False
  • fp16_full_eval: False
  • tf32: None
  • local_rank: 0
  • ddp_backend: None
  • tpu_num_cores: None
  • tpu_metrics_debug: False
  • debug: []
  • dataloader_drop_last: False
  • dataloader_num_workers: 12
  • dataloader_prefetch_factor: None
  • past_index: -1
  • disable_tqdm: False
  • remove_unused_columns: True
  • label_names: None
  • load_best_model_at_end: True
  • ignore_data_skip: False
  • fsdp: []
  • fsdp_min_num_params: 0
  • fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
  • fsdp_transformer_layer_cls_to_wrap: None
  • accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
  • deepspeed: None
  • label_smoothing_factor: 0.0
  • optim: adamw_torch
  • optim_args: None
  • adafactor: False
  • group_by_length: False
  • length_column_name: length
  • ddp_find_unused_parameters: None
  • ddp_bucket_cap_mb: None
  • ddp_broadcast_buffers: False
  • dataloader_pin_memory: True
  • dataloader_persistent_workers: False
  • skip_memory_metrics: True
  • use_legacy_prediction_loop: False
  • push_to_hub: False
  • resume_from_checkpoint: None
  • hub_model_id: None
  • hub_strategy: every_save
  • hub_private_repo: None
  • hub_always_push: False
  • gradient_checkpointing: False
  • gradient_checkpointing_kwargs: None
  • include_inputs_for_metrics: False
  • include_for_metrics: []
  • eval_do_concat_batches: True
  • fp16_backend: auto
  • push_to_hub_model_id: None
  • push_to_hub_organization: None
  • mp_parameters:
  • auto_find_batch_size: False
  • full_determinism: False
  • torchdynamo: None
  • ray_scope: last
  • ddp_timeout: 1800
  • torch_compile: False
  • torch_compile_backend: None
  • torch_compile_mode: None
  • dispatch_batches: None
  • split_batches: None
  • include_tokens_per_second: False
  • include_num_input_tokens_seen: False
  • neftune_noise_alpha: None
  • optim_target_modules: None
  • batch_eval_metrics: False
  • eval_on_start: False
  • use_liger_kernel: False
  • eval_use_gather_object: False
  • average_tokens_across_devices: False
  • prompts: None
  • batch_sampler: batch_sampler
  • multi_dataset_batch_sampler: proportional

Training Logs

Click to expand
Epoch Step Training Loss Validation Loss accuracy
0 0 - - 0.4392
0.0000 1 22.1457 - -
0.0068 200 19.4909 - -
0.0135 400 13.7828 - -
0.0203 600 9.9129 - -
0.0270 800 7.6419 - -
0.0338 1000 6.3482 - -
0.0406 1200 5.7019 - -
0.0473 1400 5.2986 - -
0.0541 1600 4.741 - -
0.0608 1800 3.7375 - -
0.0676 2000 2.7031 - -
0.0744 2200 2.0178 - -
0.0811 2400 1.6433 - -
0.0879 2600 1.3954 - -
0.0946 2800 1.2457 - -
0.1014 3000 1.1331 - -
0.1082 3200 1.0455 - -
0.1149 3400 0.9814 - -
0.1217 3600 0.9191 - -
0.1284 3800 0.8788 - -
0.1352 4000 0.8551 - -
0.1420 4200 0.8079 - -
0.1487 4400 0.7789 - -
0.1555 4600 0.7448 - -
0.1622 4800 0.7439 - -
0.1690 5000 0.7255 - -
0.1758 5200 0.6902 - -
0.1825 5400 0.6737 - -
0.1893 5600 0.6621 - -
0.1960 5800 0.6438 - -
0.2028 6000 0.6099 - -
0.2096 6200 0.6171 - -
0.2163 6400 0.5861 - -
0.2231 6600 0.5788 - -
0.2298 6800 0.5736 - -
0.2366 7000 0.5608 - -
0.2434 7200 0.5457 - -
0.2501 7400 0.5339 - -
0.2569 7600 0.5046 - -
0.2636 7800 0.513 - -
0.2704 8000 0.5087 - -
0.2771 8200 0.4826 - -
0.2839 8400 0.4836 - -
0.2907 8600 0.4761 - -
0.2974 8800 0.4655 - -
0.3042 9000 0.4421 - -
0.3109 9200 0.4436 - -
0.3177 9400 0.4418 - -
0.3245 9600 0.4272 - -
0.3312 9800 0.4334 - -
0.3380 10000 0.417 - -
0.3447 10200 0.4033 - -
0.3515 10400 0.4008 - -
0.3583 10600 0.3986 - -
0.3650 10800 0.3865 - -
0.3718 11000 0.3846 - -
0.3785 11200 0.3803 - -
0.3853 11400 0.373 - -
0.3921 11600 0.3741 - -
0.3988 11800 0.3588 - -
0.4056 12000 0.3568 - -
0.4123 12200 0.3486 - -
0.4191 12400 0.3439 - -
0.4259 12600 0.3373 - -
0.4326 12800 0.3396 - -
0.4394 13000 0.3283 - -
0.4461 13200 0.3267 - -
0.4529 13400 0.3235 - -
0.4597 13600 0.3058 - -
0.4664 13800 0.3059 - -
0.4732 14000 0.3071 - -
0.4799 14200 0.299 - -
0.4867 14400 0.2963 - -
0.4935 14600 0.2941 - -
0.5002 14800 0.286 - -
0.5070 15000 0.2894 - -
0.5137 15200 0.276 - -
0.5205 15400 0.2741 - -
0.5273 15600 0.2772 - -
0.5340 15800 0.2751 - -
0.5408 16000 0.2613 - -
0.5475 16200 0.2621 - -
0.5543 16400 0.2615 - -
0.5611 16600 0.2642 - -
0.5678 16800 0.2527 - -
0.5746 17000 0.248 - -
0.5813 17200 0.2453 - -
0.5881 17400 0.2478 - -
0.5949 17600 0.2469 - -
0.6016 17800 0.2478 - -
0.6084 18000 0.2443 - -
0.6151 18200 0.2473 - -
0.6219 18400 0.2394 - -
0.6287 18600 0.2343 - -
0.6354 18800 0.2368 - -
0.6422 19000 0.2317 - -
0.6489 19200 0.2337 - -
0.6557 19400 0.2325 - -
0.6625 19600 0.2272 - -
0.6692 19800 0.2303 - -
0.6760 20000 0.2191 - -
0 0 - - 0.5022
0.6760 20000 - 1.0284 -
0.6827 20200 0.2258 - -
0.6895 20400 0.2173 - -
0.6963 20600 0.2183 - -
0.7030 20800 0.2117 - -
0.7098 21000 0.208 - -
0.7165 21200 0.2173 - -
0.7233 21400 0.2129 - -
0.7301 21600 0.2092 - -
0.7368 21800 0.212 - -
0.7436 22000 0.2076 - -
0.7503 22200 0.2113 - -
0.7571 22400 0.2027 - -
0.7638 22600 0.2069 - -
0.7706 22800 0.2015 - -
0.7774 23000 0.2041 - -
0.7841 23200 0.2059 - -
0.7909 23400 0.2073 - -
0.7976 23600 0.2018 - -
0.8044 23800 0.2006 - -
0.8112 24000 0.1998 - -
0.8179 24200 0.2009 - -
0.8247 24400 0.1971 - -
0.8314 24600 0.1956 - -
0.8382 24800 0.1943 - -
0.8450 25000 0.1876 - -
0.8517 25200 0.2 - -
0.8585 25400 0.1892 - -
0.8652 25600 0.1923 - -
0.8720 25800 0.1879 - -
0.8788 26000 0.1874 - -
0.8855 26200 0.1868 - -
0.8923 26400 0.1853 - -
0.8990 26600 0.1869 - -
0.9058 26800 0.1879 - -
0.9126 27000 0.1862 - -
0.9193 27200 0.1891 - -
0.9261 27400 0.1793 - -
0.9328 27600 0.1786 - -
0.9396 27800 0.1736 - -
0.9464 28000 0.1787 - -
0.9531 28200 0.1874 - -
0.9599 28400 0.1798 - -
0.9666 28600 0.1818 - -
0.9734 28800 0.1799 - -
0.9802 29000 0.1763 - -
0.9869 29200 0.1758 - -
0.9937 29400 0.1799 - -
1.0004 29600 0.1737 - -
1.0072 29800 0.17 - -
1.0140 30000 0.1671 - -
1.0207 30200 0.168 - -
1.0275 30400 0.1703 - -
1.0342 30600 0.1611 - -
1.0410 30800 0.1629 - -
1.0478 31000 0.1621 - -
1.0545 31200 0.1612 - -
1.0613 31400 0.1607 - -
1.0680 31600 0.1623 - -
1.0748 31800 0.1659 - -
1.0816 32000 0.1619 - -
1.0883 32200 0.1595 - -
1.0951 32400 0.1649 - -
1.1018 32600 0.16 - -
1.1086 32800 0.1573 - -
1.1154 33000 0.1624 - -
1.1221 33200 0.159 - -
1.1289 33400 0.1596 - -
1.1356 33600 0.1573 - -
1.1424 33800 0.1544 - -
1.1492 34000 0.1512 - -
1.1559 34200 0.1592 - -
1.1627 34400 0.1602 - -
1.1694 34600 0.1562 - -
1.1762 34800 0.1531 - -
1.1830 35000 0.1504 - -
1.1897 35200 0.1533 - -
1.1965 35400 0.1578 - -
1.2032 35600 0.155 - -
1.2100 35800 0.1539 - -
1.2168 36000 0.1514 - -
1.2235 36200 0.157 - -
1.2303 36400 0.1586 - -
1.2370 36600 0.151 - -
1.2438 36800 0.1522 - -
1.2505 37000 0.1521 - -
1.2573 37200 0.1457 - -
1.2641 37400 0.1465 - -
1.2708 37600 0.1513 - -
1.2776 37800 0.1478 - -
1.2843 38000 0.1529 - -
1.2911 38200 0.1486 - -
1.2979 38400 0.149 - -
1.3046 38600 0.1516 - -
1.3114 38800 0.1486 - -
1.3181 39000 0.1423 - -
1.3249 39200 0.1479 - -
1.3317 39400 0.1499 - -
1.3384 39600 0.1455 - -
1.3452 39800 0.1482 - -
1.3519 40000 0.1471 - -
0 0 - - 0.5316
1.3519 40000 - 0.9581 -
1.3587 40200 0.1504 - -
1.3655 40400 0.1494 - -
1.3722 40600 0.147 - -
1.3790 40800 0.1457 - -
1.3857 41000 0.1494 - -
1.3925 41200 0.1428 - -
1.3993 41400 0.146 - -
1.4060 41600 0.1487 - -
1.4128 41800 0.1406 - -
1.4195 42000 0.1483 - -
1.4263 42200 0.1417 - -
1.4331 42400 0.1408 - -
1.4398 42600 0.1493 - -
1.4466 42800 0.1465 - -
1.4533 43000 0.1423 - -
1.4601 43200 0.1438 - -
1.4669 43400 0.1432 - -
1.4736 43600 0.1426 - -
1.4804 43800 0.1387 - -
1.4871 44000 0.1455 - -
1.4939 44200 0.1405 - -
1.5007 44400 0.1391 - -
1.5074 44600 0.1433 - -
1.5142 44800 0.1424 - -
1.5209 45000 0.1397 - -
1.5277 45200 0.1391 - -
1.5345 45400 0.1464 - -
1.5412 45600 0.1351 - -
1.5480 45800 0.1362 - -
1.5547 46000 0.1366 - -
1.5615 46200 0.1402 - -
1.5683 46400 0.1339 - -
1.5750 46600 0.1348 - -
1.5818 46800 0.1394 - -
1.5885 47000 0.1413 - -
1.5953 47200 0.1394 - -
1.6021 47400 0.14 - -
1.6088 47600 0.138 - -
1.6156 47800 0.1387 - -
1.6223 48000 0.14 - -
1.6291 48200 0.1372 - -
1.6359 48400 0.1379 - -
1.6426 48600 0.1332 - -
1.6494 48800 0.1341 - -
1.6561 49000 0.1319 - -
1.6629 49200 0.1363 - -
1.6697 49400 0.1327 - -
1.6764 49600 0.1297 - -
1.6832 49800 0.1345 - -
1.6899 50000 0.1362 - -
1.6967 50200 0.1323 - -
1.7035 50400 0.1351 - -
1.7102 50600 0.1334 - -
1.7170 50800 0.136 - -
1.7237 51000 0.1353 - -
1.7305 51200 0.1271 - -
1.7372 51400 0.1345 - -
1.7440 51600 0.1347 - -
1.7508 51800 0.1254 - -
1.7575 52000 0.1308 - -
1.7643 52200 0.132 - -
1.7710 52400 0.1275 - -
1.7778 52600 0.1306 - -
1.7846 52800 0.1277 - -
1.7913 53000 0.1304 - -
1.7981 53200 0.1319 - -
1.8048 53400 0.1339 - -
1.8116 53600 0.1332 - -
1.8184 53800 0.1293 - -
1.8251 54000 0.13 - -
1.8319 54200 0.1333 - -
1.8386 54400 0.1273 - -
1.8454 54600 0.1296 - -
1.8522 54800 0.125 - -
1.8589 55000 0.1332 - -
1.8657 55200 0.1288 - -
1.8724 55400 0.1233 - -
1.8792 55600 0.1282 - -
1.8860 55800 0.1341 - -
1.8927 56000 0.1275 - -
1.8995 56200 0.1286 - -
1.9062 56400 0.1279 - -
1.9130 56600 0.126 - -
1.9198 56800 0.1306 - -
1.9265 57000 0.1289 - -
1.9333 57200 0.1219 - -
1.9400 57400 0.1271 - -
1.9468 57600 0.1262 - -
1.9536 57800 0.1253 - -
1.9603 58000 0.1283 - -
1.9671 58200 0.1314 - -
1.9738 58400 0.1237 - -
1.9806 58600 0.1239 - -
1.9874 58800 0.1242 - -
1.9941 59000 0.1257 - -
2.0009 59200 0.1228 - -
2.0076 59400 0.1106 - -
2.0144 59600 0.1143 - -
2.0212 59800 0.1132 - -
2.0279 60000 0.1122 - -
0 0 - - 0.5386
2.0279 60000 - 1.0542 -
2.0347 60200 0.116 - -
2.0414 60400 0.1081 - -
2.0482 60600 0.1134 - -
2.0550 60800 0.1139 - -
2.0617 61000 0.106 - -
2.0685 61200 0.1131 - -
2.0752 61400 0.1121 - -
2.0820 61600 0.1077 - -
2.0888 61800 0.112 - -
2.0955 62000 0.1087 - -
2.1023 62200 0.1155 - -
2.1090 62400 0.1118 - -
2.1158 62600 0.1107 - -
2.1226 62800 0.1067 - -
2.1293 63000 0.1079 - -
2.1361 63200 0.1094 - -
2.1428 63400 0.1114 - -
2.1496 63600 0.1094 - -
2.1564 63800 0.1056 - -
2.1631 64000 0.1095 - -
2.1699 64200 0.1074 - -
2.1766 64400 0.107 - -
2.1834 64600 0.1124 - -
2.1902 64800 0.1121 - -
2.1969 65000 0.1115 - -
2.2037 65200 0.109 - -
2.2104 65400 0.1158 - -
2.2172 65600 0.1068 - -
2.2239 65800 0.1095 - -
2.2307 66000 0.1062 - -
2.2375 66200 0.1063 - -
2.2442 66400 0.108 - -
2.2510 66600 0.1128 - -
2.2577 66800 0.1082 - -
2.2645 67000 0.1066 - -
2.2713 67200 0.1101 - -
2.2780 67400 0.1072 - -
2.2848 67600 0.109 - -
2.2915 67800 0.1069 - -
2.2983 68000 0.1136 - -
2.3051 68200 0.1108 - -
2.3118 68400 0.109 - -
2.3186 68600 0.1139 - -
2.3253 68800 0.1081 - -
2.3321 69000 0.1082 - -
2.3389 69200 0.1064 - -
2.3456 69400 0.1064 - -
2.3524 69600 0.1084 - -
2.3591 69800 0.1058 - -
2.3659 70000 0.1109 - -
2.3727 70200 0.105 - -
2.3794 70400 0.1096 - -
2.3862 70600 0.1064 - -
2.3929 70800 0.1079 - -
2.3997 71000 0.1038 - -
2.4065 71200 0.1072 - -
2.4132 71400 0.1097 - -
2.4200 71600 0.1098 - -
2.4267 71800 0.1067 - -
2.4335 72000 0.1098 - -
2.4403 72200 0.1088 - -
2.4470 72400 0.1031 - -
2.4538 72600 0.1034 - -
2.4605 72800 0.1046 - -
2.4673 73000 0.1065 - -
2.4741 73200 0.1062 - -
2.4808 73400 0.1064 - -
2.4876 73600 0.1053 - -
2.4943 73800 0.1055 - -
2.5011 74000 0.1057 - -
2.5079 74200 0.1095 - -
2.5146 74400 0.1038 - -
2.5214 74600 0.1025 - -
2.5281 74800 0.1032 - -
2.5349 75000 0.1036 - -
2.5417 75200 0.1021 - -
2.5484 75400 0.1066 - -
2.5552 75600 0.1054 - -
2.5619 75800 0.1028 - -
2.5687 76000 0.1066 - -
2.5755 76200 0.1079 - -
2.5822 76400 0.1083 - -
2.5890 76600 0.1065 - -
2.5957 76800 0.1072 - -
2.6025 77000 0.1031 - -
2.6093 77200 0.1022 - -
2.6160 77400 0.1084 - -
2.6228 77600 0.1068 - -
2.6295 77800 0.1009 - -
2.6363 78000 0.098 - -
2.6431 78200 0.1019 - -
2.6498 78400 0.1038 - -
2.6566 78600 0.1032 - -
2.6633 78800 0.1059 - -
2.6701 79000 0.1058 - -
2.6769 79200 0.1049 - -
2.6836 79400 0.1023 - -
2.6904 79600 0.1079 - -
2.6971 79800 0.1061 - -
2.7039 80000 0.1078 - -
0 0 - - 0.5402
2.7039 80000 - 1.0363 -
2.7106 80200 0.1075 - -
2.7174 80400 0.105 - -
2.7242 80600 0.1053 - -
2.7309 80800 0.1007 - -
2.7377 81000 0.1009 - -
2.7444 81200 0.1022 - -
2.7512 81400 0.1017 - -
2.7580 81600 0.1013 - -
2.7647 81800 0.1044 - -
2.7715 82000 0.1045 - -
2.7782 82200 0.1003 - -
2.7850 82400 0.1038 - -
2.7918 82600 0.1049 - -
2.7985 82800 0.1026 - -
2.8053 83000 0.1082 - -
2.8120 83200 0.1033 - -
2.8188 83400 0.1044 - -
2.8256 83600 0.1013 - -
2.8323 83800 0.1057 - -
2.8391 84000 0.0992 - -
2.8458 84200 0.0973 - -
2.8526 84400 0.102 - -
2.8594 84600 0.1048 - -
2.8661 84800 0.1013 - -
2.8729 85000 0.1038 - -
2.8796 85200 0.1027 - -
2.8864 85400 0.1037 - -
2.8932 85600 0.1007 - -
2.8999 85800 0.1033 - -
2.9067 86000 0.1016 - -
2.9134 86200 0.1051 - -
2.9202 86400 0.1001 - -
2.9270 86600 0.1003 - -
2.9337 86800 0.0969 - -
2.9405 87000 0.1056 - -
2.9472 87200 0.0997 - -
2.9540 87400 0.1 - -
2.9608 87600 0.1008 - -
2.9675 87800 0.1003 - -
2.9743 88000 0.0991 - -
2.9810 88200 0.1022 - -
2.9878 88400 0.1012 - -
2.9946 88600 0.0992 - -
3.0013 88800 0.1002 - -
3.0081 89000 0.0885 - -
3.0148 89200 0.0918 - -
3.0216 89400 0.0888 - -
3.0284 89600 0.088 - -
3.0351 89800 0.0893 - -
3.0419 90000 0.0896 - -
3.0486 90200 0.0899 - -
3.0554 90400 0.0918 - -
3.0622 90600 0.0888 - -
3.0689 90800 0.0961 - -
3.0757 91000 0.0867 - -
3.0824 91200 0.0927 - -
3.0892 91400 0.0891 - -
3.0960 91600 0.0864 - -
3.1027 91800 0.0884 - -
3.1095 92000 0.0879 - -
3.1162 92200 0.088 - -
3.1230 92400 0.086 - -
3.1298 92600 0.0885 - -
3.1365 92800 0.0918 - -
3.1433 93000 0.0885 - -
3.1500 93200 0.0873 - -
3.1568 93400 0.0922 - -
3.1636 93600 0.0878 - -
3.1703 93800 0.0893 - -
3.1771 94000 0.0865 - -
3.1838 94200 0.0888 - -
3.1906 94400 0.0871 - -
3.1974 94600 0.0884 - -
3.2041 94800 0.0858 - -
3.2109 95000 0.0872 - -
3.2176 95200 0.0871 - -
3.2244 95400 0.0887 - -
3.2311 95600 0.0903 - -
3.2379 95800 0.0865 - -
3.2447 96000 0.0901 - -
3.2514 96200 0.0894 - -
3.2582 96400 0.0945 - -
3.2649 96600 0.0851 - -
3.2717 96800 0.086 - -
3.2785 97000 0.0884 - -
3.2852 97200 0.0889 - -
3.2920 97400 0.0878 - -
3.2987 97600 0.0872 - -
3.3055 97800 0.0885 - -
3.3123 98000 0.0861 - -
3.3190 98200 0.0854 - -
3.3258 98400 0.0871 - -
3.3325 98600 0.0884 - -
3.3393 98800 0.0833 - -
3.3461 99000 0.0912 - -
3.3528 99200 0.0859 - -
3.3596 99400 0.088 - -
3.3663 99600 0.0861 - -
3.3731 99800 0.0908 - -
3.3799 100000 0.0852 - -
0 0 - - 0.5332
3.3799 100000 - 1.1603 -

Framework Versions

  • Python: 3.11.0
  • Sentence Transformers: 4.0.1
  • PyLate: 1.1.7
  • Transformers: 4.48.2
  • PyTorch: 2.6.0+cu124
  • Accelerate: 1.6.0
  • Datasets: 3.5.0
  • Tokenizers: 0.21.1

Citation

BibTeX

Sentence Transformers 

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084"
}

PyLate 

@misc{PyLate,
title={PyLate: Flexible Training and Retrieval for Late Interaction Models},
author={Chaffin, Antoine and Sourty, Raphaël},
url={https://github.com/lightonai/pylate},
year={2024}
}